In the 1993 AFC wild-card game, the Buffalo Bills mounted the greatest comeback in NFL history, posting a 41–38 win over the Houston Oilers after trailing by thirty-two points just a few minutes into the third quarter. The Oilers had been up 28–3 in the first half, and the Bills outscored them 35–3 after the last Oilers touchdown early in the third quarter to pull out the victory. A one-yard run for a Buffalo touchdown followed by a successful onside kick marked the beginning of the comeback.1 It would be tempting to explain the dramatic turnaround in this game as the result of a shift in momentum, possibly started by the touchdown and the recovered onside kick. Certainly the game contained what would appear to be two different demonstrations of momentum—the Oilers dominated the first half, whereas the Bills rolled to victory in the second—and we can say that there was a shift in the patterns of scoring. Is momentum indeed part of the explanation of what happened in this historic game?
The typical discussion of momentum in football goes something like this: “Coach Jones would sure like to make this field goal before halftime so the Panthers can go into the locker room with the momentum on their side.” Or: “Well, we've just seen a pretty dramatic shift in the momentum of the game after the Owls coughed up that fumble in the last series, Coach.” But what is momentum? It seems to be something that can shift. It seems to be something that can be conserved between games and can perhaps be conserved even between the end of one season and the beginning of the next. Does it refer to anything real at all? Does it, instead, refer to an event that normally would be described in other terms? Can momentum be more than a matter of playing well or of having the breaks go your way in a football game?
Momentum and Success
Whatever else it refers to, the idea of momentum is often used as an explanation of things going well on the field. We never, or almost never, appeal to the idea of momentum when things are going badly. It would be very strange to hear an announcer refer to momentum during a stretch of bad play—dropped passes, missed blocks, interceptions. This much is obvious, but it is just this feature of momentum-talk that should make us suspicious, at least if we would like momentum to do some explanatory work. Why should this be so? The problem is that if momentum is going to explain successful football playing—if it is to be something other than successful football playing—then we need a way of identifying momentum that is independent of the evidence offered for its presence. So it would be good if we could find some way of separating momentum from the evidence for momentum, and doing this seems to be difficult. And there is another feature that complicates things for us: why is it that, in common usage, only one football team in a given game can exhibit momentum at a time? At the very least, this fact about the way that we use this term should give us pause, for if momentum is a separate phenomenon, if it refers to something more than success itself, then why couldn't two teams have it at the same time? We could imagine two teams in a well-played, hard-fought game playing to a standstill, but no one would think that the two teams in this imagined situation were exhibiting momentum. It would seem that one team can have momentum only if the other team lacks it, and so we have another piece of our puzzle: a team is said to have momentum when it is playing well and the other team is not. Are there other components to this concept? Perhaps, but let us see whether we can make sense of what we've got.
The Mystery of Momentum
It is possible that talking about momentum is nothing more than a shorthand way of pointing out that one team is playing well while its opponent is not. If this is true, then we can forgive announcers and others who routinely discuss momentum and find them innocent of the charge of appealing to mysterious or even superstitious explanations. Of course coaches and players want momentum, on this view, since having momentum would be the same thing as playing well while your opponent does not. But a careful look at how we actually use this concept seems to point to something beyond this alternative way of expressing the obvious. It seems that the concept of momentum is intended to explain something and that the concept means more than simply playing well. Let's try again.
It might be that some mysterious force or power is being pressed into service here. How literally should we take the claim that the Bills are playing well because they have the momentum? True, we do talk about momentum shifting and about teams gaining and regaining momentum, and this might mean that momentum is a something that can be gained and lost. Now, not every noun indicates the existence of a thing, and perhaps momentum is like, say, talent: talent is not a something in the way that a can of beer is a something, but the term might refer to a condition or a potential or a power. Is momentum like this?
The idea of talent is promising here, because it shares another feature with momentum: having talent is also indistinguishable, mostly, from the evidence for talent. Even so, most of us talk freely about artists and athletes as though they either are or are not talented. So maybe talent and momentum will stand or fall together: either we can use both with good conscience or we can use neither. But does the concept of talent explain anything? To be sure, talent is not absolutely indistinguishable from the evidence for talent. We do talk about persons who squander their talent, and this seems to open up enough daylight between the concept and the evidence for the concept so that perhaps the concept does some work. But even here, when we talk about talent that is wasted, we are generally basing our belief in that talent on a demonstrated ability to do something, and so there isn't that much daylight, perhaps, after all. And the situation is worse for momentum, because it is hard to imagine the existence of any daylight between the concept of momentum and the evidence for momentum.
Still, every football player I have talked with about this subject has maintained a devout belief in the existence of momentum, and it seems mean-spirited to dismiss these players’ beliefs so easily. Could it be that momentum refers to a complex but real phenomenon that is shown in successful performance but that refers to something more than successful performance itself?
One key element in the idea of momentum is that success leads to continued success. Momentum on this psychological account would refer to successful performance that is at least partly explained by confidence on the part of players. To this we might add a corresponding lessening of confidence on the part of the opponents: as one team believes in itself, the other begins to have doubts about its ability to withstand the efforts of the other. The “confidence” explanation seems to make sense of other features of the momentum phenomenon, the fact that momentum can come and go and that we don't normally attribute momentum to both teams at once. It stands to reason that as the deficit between the Bills and the Oilers began to shrink, the Bills would gain confidence and the Oilers would lose it. Additionally, the confidence explanation helps us to make sense of the way in which shifts in momentum seem to have an element of randomness; they occur as if dispensed as a matter of grace and not of desert.
What empirical support is there for the hypothesis that confidence causes success?
Though there are numerous explanations for the phenomenon of momentum and a growing body of research, no scientific consensus has been reached as to the preferred explanatory model to be used, and the actual empirical research to date is less than conclusive.
A Model for Understanding Momentum
Jim Taylor and Andrew Demick propose a theoretical model of momentum that uses the idea of a “momentum chain,” a series of changes that result in momentum:
a. Precipitating event or events
b. Change in cognition, affect, and physiology
c. Change in behavior
d. The resulting increase or decrease in performance consistent with the above changes
e. A contiguous and opposing change in the previous factors on the part of the opponent (for sports with head-to-head competition)
f. A resultant change in the immediate outcome.2
Item (b) would include the confidence that results from a precipitating event, so this model clearly makes room for a psychological explanation of momentum. One benefit of this model is that it explains why momentum fails to occur even when we might expect it to following an appropriate precipitating event: Taylor and Demick suggest that the disconnect between expected momentum shifts and actual outcomes is the result of failure at different links of the chain. To show that there is in fact a correlation between precipitating events and performance outcomes, they conducted two preliminary studies. One study involved basketball, and the other involved tennis. Interestingly, though their model predicts that momentum will be less likely to occur in a team sport because individual players will traverse the momentum chain at different rates of speed, it was in basketball that they were able to show that precipitating effects were important for the production of changes in immediate outcomes. In tennis, by contrast, no statistically significant differences were observed between the presence and absence of precipitating events and immediate outcomes.3
David Romer has created a stir recently by publishing an analysis that, if correct, shows that NFL coaches are far too careful when it comes to punting on fourth down, and his analysis has something to say about momentum, too. Romer found that the expected average advantage of going for a first down on fourth down is greater than the expected disadvantage. So in the long run, a coach who always goes for it on fourth down will win more games than he would have won if he had punted most of the time and gone for it only occasionally.4 One objection that Romer considers is whether his analysis is undercut by the possibility of a team's losing momentum if they go for it on fourth down and fail. It might turn out that the expected advantage of always going for it won't materialize because his argument has not taken into consideration the likelihood of the damaging effects of being on the losing end of a momentum shift. So Romer provides a rebuttal to this potential objection, and his response is simple: momentum does not exist.5
Romer's methodology resembles that of other studies of momentum. He looks at expected outcomes following very good or very bad plays to see whether there is a tendency for a good play to generate further good plays or a bad play to generate further bad plays. He finds no such statistically significant tendency. In fact, there is a slight tendency for teams that have endured very bad plays (which in this case means they have lost possession of the ball) to do somewhat better than average in the situation that follows the bad play.6
So far the evidence of the existence of momentum is equivocal. Some studies are able to make an impressive case for momentum, though the definitions of course vary somewhat, and others appear to show that any correlations between events that jump-start momentum and subsequent performance can easily be explained as occurring by chance. Still, in some cases at least, the evidence for momentum, no matter what the model, demands a second look.
One difficulty in developing a model with testable predictive value is that it is difficult to include the subjective experiences of athletes in real time. We might interview athletes during games, but that presents a number of challenges, the most obvious being that players will be unable to stop play so that they can answer questions. Physiological monitoring shows more promise, and Taylor and Demick's model already contains the physiological as a link in the chain. But why are the subjective experiences of athletes important?
Momentum under the Helmet
We need to get at the subjective experience of athletes because, intuitively, momentum is thought to be a feeling of mastery and success at achieving athletic goals. The psychological models of momentum seek to link the experience of mastery and success with previous and current performance, and if we removed the subjective side of things, we would be left with no obvious causal mechanism for the linkage of present and future success. Subjective experience without actual positive performance would not be momentum, and a mere linkage of earlier and later positive performance without appeal to the subjective experience of mastery and success might be of little value, for at least two reasons. First, athletes hope to be able to understand momentum and use it. A number of sports advisers purport to be able to teach athletes how to initiate positive “jump starts” or to take advantage of positive events when they happen randomly. If momentum can't be manipulated, then athletes won't be able to profit from studying it. Second, correlations that make no use of the athletes’ conscious responses to events during a game just seem plain uninteresting. Whatever the connection might be, if there is no causal connection that runs through the athletes’ own conscious experience, it is not momentum.
What we need here is a way to access a team's collective sense of momentum (in the case of football) and a way of marking the initiation or change of momentum; we need a way of defining a momentum-initiating event. But why should this be the case? Couldn't we say that a team might enter a game with the requisite feelings of mastery and success and then go on to dominate its opponent with ease? The many one-sided Super Bowls in the history of the NFL serve as examples of this possibility. Why wouldn't these be cases of momentum? And how would we recognize the difference between a team's beginning a game with solid play because it is a superior team and a team's entering a game and getting a lucky break that, because of the power of momentum, it then turns into a dominant performance?
Most of the studies seem to think of momentum as something whose existence begins as a result of a particular episode, but how should we think of the early episodes of dominant teams in games? Are we to say that excellent teams always or nearly always exhibit momentum? Excellent teams certainly seem to fulfill most if not all of the requirements sometimes listed for momentum: a sense of confidence, mastery, flow, and success. Should momentum require that a particular event cause or intensify these qualities? But if momentum ends up being displayed in just any performance that exhibits these characteristics, then we run the risk of using momentum to say no more than that a team is playing well. There is some appeal, then, in the idea that momentum requires an initiating event.
Is Momentum a Metaphysical Reality?
It might seem that if there is at least some case for the existence of momentum, then use of the idea of momentum by athletes, sportscasters, and fans is also justified. But this isn't necessarily so. It all depends on how large the momentum effect might be. It might turn out that, though there is statistical reason for thinking that success does breed success, the real question is how much momentum explains relative to other factors, including sheer randomness. Given the uncertain nature of the studies done on momentum so far, and the relatively small momentum effects claimed, it appears that momentum, especially in some sports, will turn out to be less likely as an explanation than the combined effects of other explanations—chance, principally—and even less likely than other single causes. So perhaps there is some boost in success rates that occurs as a result of athletes’ responding well to precipitating events; even so, unless that success is greater than the studies to date seem to support, then we will usually be wrong when we refer to momentum as a significant explanation. The situation is like cigarette smoking: we know that smoking increases your chance of having a heart attack, but we are unable in any particular case to say that smoking was the cause, since many other factors are generally at work as well.
So, even if momentum is real, its appeal is greatly overrated, given its relative unimportance as an explanation for athletic success. The amount of success caused by momentum—even according to the studies that support its existence—is usually small in comparison with the sum total attributable to the other causal factors involved, so appeals to momentum as explanations of a team's success will often be unjustifiable. Since the best evidence for momentum is almost always going to involve statistics unavailable to the casual observer, appealing to momentum will be like the wife of a victim appealing to smoking as a cause of her husband's heart attack. She may feel that she just knows that smoking did it, but she will not be in a position to establish her case in any plausible way.
Let us sum up the points made so far. In at least a few sports, momentum would appear to be real. There appears to be a phenomenon in which success makes it more likely that success will follow, though even here there are questions about definition: does momentum as measured by social scientists match its usage by athletes, sportscasters, and fans? In any given case, the evidence—typically statistical in nature—needed to justify an attribution of momentum in a football game will be lacking, so that any appeal to that concept as an explanation of a particular shift in the fortunes of a particular team will be suspect.
Consider, now, a typical scenario. The Tigers score a touchdown on a thirty-yard pass and take a narrow lead of 10 to 7. They then score another touchdown just before the half. In addition, their fans and perhaps the sportscasters feel that they can see improved body language in the players: they are hustling, they're in a groove, they are not getting procedure penalties, they're protecting the quarterback, and so on. Things are going their way and they seem to be confident, though earlier in the game they appeared hesitant and out of sync somehow. Now, do the Tigers have the momentum going into the locker room? Further, suppose that researchers have reached a consensus that, for a well-defined list of momentum initiators, a team receives a 20 percent momentum premium. In other words, suppose that a team is 20 percent more likely to score if the players have been jump-started with one of the items on the list—recovered fumble, interception, and so on—than if they had simply received a kickoff (since just having the ball in itself makes it more likely that a team will score, obviously). Would we be justified in saying that the Tigers have momentum?
Assume that the “evidence” is appropriate; that is, assume that what we are observing in this case is just what our ideal researchers say we should observe if we hope to observe momentum as described in someone's ideal model of what momentum is. In other words, this would be an ideal case: we are observing the kind of event that, when observed hundreds of times, say, would be expected to yield a statistically significant momentum effect. In this case, can we say that the Tigers have momentum?
Unfortunately, my purely hypothetical 20 percent momentum premium would represent only an average, and we would expect it to be spread out in an unpredictable way through most games. Perhaps in some games momentum accounts for more than 20 percent of scoring, and in other games it accounts for less. This means that, though momentum might on occasion be a significant explanation for the success of a football team, in general we will not be in a position to say that it is a significant factor. Given the size of the momentum effects we would expect to discover statistically, we would be well advised to be cautious in appealing to momentum in any given case.
Conclusion: Don't Punt on Fourth Down
Finally, let us return to the possibility that David Romer is right in saying that momentum plays no role in football. If he is right, then it is important for coaches and players to know that. Romer's conclusion about punting on fourth down is a good example of how our belief in momentum may lead us to play too conservatively, since we will factor in the possibility of losing momentum if we fail to make a first down. Removing the threat of a momentum loss might also enable players of other sports to be more aggressive in situations that might not obviously justify taking risks. Paradoxically, if Romer is right and we would win more games by playing more aggressively, then our belief in momentum and the possibility of losing it might actually lead us to lose games that we would otherwise have won.
Notes
1. Pro Football Hall of Fame, The NFL's Greatest Comeback, http://www.profootballhof.com/history/.
2. Jim Taylor and Andrew Demick, “A Multidimensional Model of Momentum in Sports,” Journal of Applied Sport Psychology 6 (1994): 51–70, http://www.drjimtaylor.com/articles/scholarly.html.
3. Taylor and Demick, “Multidimensional Model,” 51–70.
4. David Romer, “Do Firms Maximize? Evidence from Professional Football,” Journal of Political Economy 114, no. 2 (April 2006): 342, http://find.galegroup.com.
5. Romer, “Do Firms Maximize?” 358.
6. Romer, “Do Firms Maximize?” 358.